Implementing GF Resource Grammar for Sindhi
language
A subtitle that can be quite long if necessary
Master of science Thesis in Intelligent System Design
JHERNA DEVI OAD
Department of Applied Information Technology
CHALMERS UNIVERSITY OF TECHNOLOGY
Gothenburg, Sweden, 2009
Report No. 2012:070
ISSN: 1651-4769
Implementing a Grammatical Framework Resource grammar for Sindhi
Abstract
Recently, demand for getting data through the world wide web has been dramatically
increased. People with different languages can find information in languages which are varied
with their own. But if we can manage somehow to translate information from one language to
another language, then it might become possible to use the complete information.
Grammatical Framework makes it easier to translate text from one language to another.
Grammatical Framework is a functional programming language that can be used to develop
multilingual natural language processing applications. Grammatical Framework programs
implement the grammatical rules of natural language. The resource grammar library is an
important part of the Grammatical Framework. The multilingual Grammatical Framework
resource library contains a language-independent abstract syntax and a set of language-
dependent concrete syntaxes.
The main focus of this thesis work is to develop a resource grammar of Sindhi language using
Grammatical Framework. This work tries to cover in detail the morphology and the syntactic
structure of Sindhi language. The developed Sindhi resource grammar can be used for
multilingual applications, such as translation, language teaching and software localization.
Acknowledgement
I wish to express my sincere gratitude to Prof. Aarne Ranta for giving me this opportunity to
work on such an interesting project. I am heartily thankful to my supervisor Shafqat Virk,
whose encouragement, supervision and support from the preliminary to the concluding level
enabled me to develop an understanding of the Grammatical Framework. Also my
appreciation goes to Mutee-u-Rehman Assistant Professor at department of computer science
Isra University Hyderabad Sindh Pakistan, his high quality research on Sindhi
grammar has made an easy understanding and formalizing a resource grammar of Sindhi
language in GF. I would also like to mention that my husband Jagdesh kumar masters student
at RWTH Aachen Germany helped me to improve the quality of the thesis.
Finally I would like to give a very special acknowledgement and gratitude to my father
Ghansho Mal, my mother Sundri Bai, my siblings and my child, who encouraged and
supported me to accomplish this goal.
ii
Table of Contents
Abstract.................................................................................................................................. ii
Acknowledgement.................................................................................................................. iiiChapter 1: Introduction........................................................................................................ 1
1.1 Linguistics............................................................................................................... 11.2 Grammatical Framework....................................................................................... 1
1.2.1 Resource Grammar.................................................................................... 2
1.2.2 Application Grammar................................................................................ 21.2.3 Indo-Iranian Languages in GF Resource Grammar Liabrary.................... 3
1.3 Sindhi Linguistics.................................................................................................... 4
1.3.1 Sindhi Script............................................................................................... 4Chapter 2: Sindhi Morphology............................................................................................. 5
2.1 Lexicon.................................................................................................................... 5
2.1.1 Nouns......................................................................................................... 52.1.2 Verbs.......................................................................................................... 8
2.1.3 Adjectives.................................................................................................. 102.1.4 Numerals.................................................................................................... 11
2.1.5 Adverbs...................................................................................................... 122.1.6 Pronouns.................................................................................................... 13
2.1.7 Conjunctions.............................................................................................. 14
2.1.7.1 Coordinate Conjunctions.......................................................... 142.1.7.2 Subordinate Conjunctions......................................................... 15
2.1.8 Interjunctions............................................................................................. 15Chapter 3:Syntax................................................................................................................... 16
3.1 Analytical Syntax.................................................................................................... 16
3.1.1 Noun Phrase (NP) ..................................................................................... 163.1.2 Verb Phrase (VP) ...................................................................................... 18
3.1.3 Adjectival Phrase (AP) ............................................................................. 213.1.4 Adverbs...................................................................................................... 22
3.1.5 Copula (Auxiliary Verb) ........................................................................... 22
3.2 Synthetical Syntax................................................................................................... 233.2.1 Clauses....................................................................................................... 23
3.2.2 Sentences.................................................................................................... 25Chapter 4:Applications......................................................................................................... 26
Chapter 5:Evaluations........................................................................................................... 30
Chapter 6:Conclusion............................................................................................................ 30Chapter 7:Future Work........................................................................................................ 31References............................................................................................................................... 32Appendix................................................................................................................................. 34
iii
Chapter 1. Introduction
This thesis project implements Sindhi computational grammar in Grammatical Framework
that can be used for text translation in multilingual grammars that has been already
implemented in Grammatical Framework . There was not any particular translation tool that
could provide the facility for text translation into Sindhi language. Grammatical Framework
resource grammar for Sindhi is the one which provides text translation into Sindhi language.
1.1 Linguistics
In this modern world human wants to share and access the knowledge but due to language
difference Its not convenient to retrieve related information from a resource.
However, In order to understand once idea and thoughts, there is a need to translate one
language to another language. It is very important to understand the grammar of both
languages including morphology and syntax. linguistically every language has an appropriate
structure including Phonetics, Morphology and syntax. In the 1950, United States originated
computational linguistics when computer started to translate Russian text into English [15].
After that computational linguistics started to grow rapidly by developing different
applications that provided such facility.
It was very important to develop a central system or tool that could provide facility of
multilingual translation. The google has provided such facility but it is limited and it does not
cover most of the world languages. Grammatical Framework provides such opportunity and
it makes easier to implement different language resource grammars.
1.2 Grammatical Framework
Grammatical Framework (GF, Ranta, 2004) is a functional and natural language processing
programming language. GF is designed for writting grammars and GF programs are based on
the Grammatical rules of the language. It is based on Martin-Löfs type -theory (Martin -Löf,
1975) and implemented in Haskell.
It is a structure that manages and defines grammars [13]. Generally GF has a module system
consisting of two main modules, one is an abstract syntax and another is concrete syntax. GF
consists of a central data structure known as abstract syntax which describes the semantic
concept. The concrete syntax that makes possible to describe how abstract syntax rules can
be linearized in to different natural language. Grammatical Framework can parse and
translate text into multiple languages simultaneously [5]. GF allows to have more than one
concrete syntax for a common abstract simultaneously and results into multilingual grammars.
Logical representation of a grammar is an abstract syntax which is independent of any
language. It defines a list of different lexical (i.e. Adjectives, verbs, nouns,adverbs, etc.) and
syntactical (Adjective phrase, noun phrase, verb phrase adverb phrase, etc.) categories (cat). It
also defines a set of tree structure rules (fun). The concrete syntax depends on natural
languages. It allocates a linearization (lin) to each function and linearization type (lincat) to
each category of the abstract syntax.
GF has a library known as Grammatical frameowrk resource library covering the
morphology and syntax of natural languages. Resource library of Grammatical Framework is
used for natural language processing . The Grammatical Framework libraries are of two
types, Resource grammar and application grammar. At present GF contains 27 resource
grammar libraries of different languages.
1.2.1 Resource Grammar
The Grammatical Framework resource grammar library consist of a set of natural language
grammar which is implemented in Grammatical Framework. These grammars are built upon
a common tree structure or a common abstract syntax. As a resource for language processing
assignments (natural language interfaces, software localization, translation etc) the resource
library can be used. The Resource Grammar Library is categorised roughly into two parts,
morphological and syntactic. It defines the morphology and comprehensive part of the syntax
of languages present in GF. As it has common abstract syntax which make convenient to
implement another new language. The resource library provides an opportunity that allows
non-linguist application programmers can write multilingual application programs easily [14].
1.2.2 Application Grammar
These grammars are domain specific grammars which encode semantic construction. The
domain experts are supposed to write application grammars. Application grammar consists of
an abstract syntax that defines the semantics of an application domain linguistically. The
resource grammar has abstract syntax that defines the linguistic structure. The application
grammar has a concrete syntax that can be expressed as a mapping to the abstract syntax. The
application grammars encode the semantics of the particular domain by using resource
grammar through abstract interfaces [3].
2
1.2.3 Indo-Iranian Languages in Gramatical Framework Resource Grammar Liabrary
Grammatical Framework has been used to implement grammars of a number of natural
languages. Indo-Iranian languages in Grammatical Framework resource grammar liabrary
includes Urdu, Punjabi, Sindhi. These languages can not use common code. Every language
need to implement seperate code.
The implemented Sindhi resource grammar using Arabic script like Urdu and Punjabi. Urdu
and Punjabi languages consist of 38 alphabets whereas Sindhi alphabet consists of 52
alphabet characters and two special characters. The direction of the writing of Sindhi, Urdu
and panajbi is right-left.
Aforementioned languages have common scripts, but few similar alphabet characters. These
languages have different grammars. All language has its own grammar rules.
Sindhi noun phrase has 11 case makers and noun phrase is very important part of syntax of
the language. Coupla differentiate the tense of sentence, in Sindhi language sentences are
incomplete without coupla. Urdu syntax using ezafe constructions like as Persian language.
There is not any concept of ezafe construction in Sindhi language.
Language Script IPA
English Thank you
Sindhi رباني انجي م ھتو ھ tohaːnɟi: mhrba nː i:
Urdu :A:pK a: shukrīy a پکا شکریا ا
Punjabi ا شکریہ ڈتہا tuhā āɖ shukrīāTable 1.2.3: Sindhi ,Urdu and Punjabi script
3
1.3 Sindhi Linguistics
The main objective of this work is to implement GF resource grammar for Sindhi language.
Sindhi falls under an Indo-Aryan Group of the Indo-Iranian Language family.
Sindhi language is one of the South Asian regional languages, specially the language of Sindh
province of Pakistan. Sindhi language is one of the most widely used language of almost 78
million speakers including 41 million speakers in Pakistan, 12 and 25 million respectively
from India and other countries.
1.3.1 Sindhi Script
The Sindhi language is written in two scripts, Arabic and Devanagari script.
Arabic script consists of 52 alphabet characters known Arabic-Sindhi script, standardized by
British government in 1852. It is written from right to left, this script is used by people of
Pakistan, where as Sindhi spoken in India has Devanagari script from left to right with
additional four letters which express the special implosive sounds. Sindhi language can also
be written with characters of the Gurmukhi as well as Gujrati alphabet.
The major characteristics of Sindhi are Rich morphology, verb-compounding, relatively free
word-order, which inflects and derivates using suffixes and prefixes [8].
ھج ڄ ج پ ث ٺ ٽ ٿ ت ڀ ٻ ب ا
ɟʱ ʄ ɟ p S t̪ʰ t̪ tʰ t bʱ ɓ b aː
ڙ ر ذ ڍ ڊ ڏ ڌ د خ ح ڇ چ ڃ
ɽ r Ð ɖʱ ɖ ɗ dʱ d x ħ cʰ c ɲ
ق ڦ ف غ ع ظ ط ض ص ش س ز ء
K pʰ F ɣ ʕ ʐ ʈ z ʂ ʃ s z ʔ
ي ھ و ڻ ن م ل ڱ ھگ ڳ گ ک ڪ
iː h O ɳ N m l ŋ ɡʱ ɠ ɡ kʰ k
Table 1.1: Sindhi Arbic script and IPA (International Phonetic Alphabate)
4
Chapter 2. Sindhi Morphology
Sindhi language has a very rich morphology, inflection and derivations of the script treated by
prefixes and suffixes. It has a large amount of morphological variants. There are primary and
secondary words present in Sindhi language. The primary words are indivisible but the
secondary words are divisible into compound (combination of two or more primary words)
and complex words (addition of prefixes or suffixes) [8]. Morphologically Sindhi language
uses reflexive (passive) and objective (impersonal) verbs trunks (Stems), also uses suffixes
pronouns along with nouns, post-positions and verbs. [1]
2.1 Lexicon
A lexicon contains word entries and language rules. The words are stored in the lexicon
according to the occurrence of parts of speech in Sindhi language. In the Sindhi lexicon the
inflectional forms and meaning of a word is stored [12].
The Sindhi lexicon consists of two types of words, one containing simple words which has
single meaningful unit and the other, covering complex words which can be divided in more
than one meaningful units. A single meaningful unit is known as morpheme.
The words consist of more than one morpheme can be predominate in the language. These
words have one root which constitutes core of the word and one or more units as modifiers
known as affixes.
The GF resource grammar has a test lexicon. Test lexicon is built through a lexical function
known as lexical paradigms; (Bringert et. al 2011). The paradigms are used to make an
inflection table containing possible forms of the word according to lexical rules of the
language. Test lexicon consists of 450 words with different lexical categories. These lexical
categories are selected for testing randomly [9].
2.1.1 Nouns
Sindhi nouns may be simple or complex words. Nouns are classified Grammatical ly by
number (singular and plural), case (Direct, oblique, ablative and vocative) and gender
(masculine and feminine). In Sindhi language gender is classified in to two categories one is
animate noun and an other is non-animate noun as in Urdu language. Animate nouns are
specified with gender inherently where as non-animate nouns are not specified with gender
inherently but it is specified artificially.
5
Cases /Masculine (Boy ,رو ڪو okərʧ _ڇ ʊʰ ) Singular Plural
Direct (Dir) رو ڪو _ڇ okəroʧʰ را ڪو _ڇ ok rʧ ɪ ɑʰ
Oblique (Obl) ري ڪو _ڇ ok reʧ ɪʰ رن ڪو _ڇ ok rnʧ ɪʰ
Vocative (Voc) را ڪو _ڇ ok rʧ ɪ ɑʰ را ڪو _ڇ ok rʧ ɪ ɑʰ
Ablative (Abl) ري ڪو _ڇ ok reʧ ɪʰ رن ڪو _ڇ ok rn ʧ ɪʰ
Table 2.1: Inflectional forms of noun okəro_ʧʰ boy.
GF stores different inflectional forms of noun and lexical details in a record-type structure.
The construction of the noun in GF is:
Cat N;
lin Noun = {s : Number => Case => Str ; g : Gender};
It shows a record of a noun having two fields ‘s’ and ‘g’, where as ‘s’ explains ‘s :
Number => Case => Str’ (a table from number to case to string). The entity s is a
parameter to which number and case are the assigned values. The parameters number
and case are given here:
Param number = sg | Pl;
Param Case = Dir | Obl | Voc | Abl;
As mentioned above that GF stores different inflectional forms of noun and lexical details in a
record-type structure. So, ‘s’ is forming an inflection table and storing different forms of
noun inflecting for gender, case and number, ‘g’ is the type of ‘Gender’ represents that the
noun is feminine or masculine.
Parameter number inflects for singular (sg) and plural (Pl). Where as parameter case
inflects for directive (Dir), oblique (Obl), ablative (Abl) and vocative (Voc).
As its been mentioned, there is a Grammatical function (fun) list and a categories
(cat) list in an abstract syntax. The Grammatical functions (fun) declare words of
lexical categories as constant. The linearization function (lin) and linearization type
6
(lincat) to each category are assigned by concrete syntax. As an example consider the
following abstract syntax and concrete syntax code for implementation of the noun
‘boy’.
fun boy_N : N
lin boy_N = mkN01 " رو ڪو okəro ;ʧʰ , "ڇ
In GF the abstract syntax code (fun) defines that boy is of type ‘N’ (noun), while its
linearization rule ( lin ) uses lexical paradigm mkN01 and builds all inflection forms
of noun. In the above example ‘mkN01’ takes argument‘رو ڪو ) ’ڇ okəro ) and buildsʧʰ
an inflection table. The definition of function mkN01 is given as:
Oper mkN01: Str -> Noun;
Oper mkN : (x1,_,_,_,_,_,_,x8 : Str) -> Gender -> Noun =
\sd,so,sv,sa, pd,po,pv,pa, g -> {
s = table {
Sg => table { Dir => sd ; Obl => so ; Voc => sv ; Abl => sa
} ;
Pl => table { Dir => pd ; Obl => po ; Voc => pv ; Abl => pa
}
} ;
g = g
} ;
mkN01 chokro = let chokr = (tk 1 chokro)
in mkN (chokro) (chokr+"y") (chokr+"a") (chokr+"y")
(chokr+"a") (chokr+"n") (chokr+"a") (chokr+"n")
Masc ;
Here ‘chokro’ is as variable which takes a noun then creates its root and adds different
information related to the noun to make its inflectional forms.
The above function builds record as mentioned bellow.
7
S . Sg => Dir => رو ڪو okəroʧʰ -- ڇ
S . Sg => Obl => ري ڪو ڇ -- ok reʧ ɪʰ
S . Sg => Voc => را ڪو ڇ -- ok rʧ ɪ ɑʰ
S . Sg => Ab l => ري ڪو ڇ -- ok reʧ ɪʰ
It is noted that in the record ‘S’ the inflectional forms of a noun ‘boy’ are stored
which are inflecting for case, number and inherent gender. Since there are 14 lexical
paradigms which cover all groups of Sindhi nouns, however here in this example only
one lexical paradigm of noun is explained.
2.1.2 Verbs
Sindhi verbs have a complex structure. These verbs inflect for aspect (perfective,
imperfective), mood (subjunctive, imperative), gender (Masculine, Feminine), tense (Present,
Past, Future), number (Singular, Plural) and person (first, second casual, second respect, third
near, third distance). Sindhi verbs have single class for all verbs, inflect alike but with minor
difference. There are total sixty forms of the verb which have been implemented in GF Sindhi
resouce grammar library. The following structure builds table and stores different forms of
verbs in GF.
cat verb;
lin verb = {s : VForm => Str} ;
Here ‘s’ is inflectional table storing all the inflectional forms of the verb from VForm to
String. The table has only one field s. The parameter of the ‘VForm’ is defined as follows:
Param VerbForm = VF VTense PPerson Number Gender
| Inf | Inf_Fem | Inf_Obl | Ablative
| Root;
The parameter VerbForm has constructors, its use and description is defined as
VF stores different inflecting forms of the verb which inflect for tense, number, person and
gender. The VTense, PPerson, number and gender are known as context parameters and they
are explained as below.
8
param
Gender = Masc | Fem ;
VTense = Subj | Perf | Imperf;
PPerson = Pers1
| Pers2_Casual
| Pers2_Respect
| Pers3_Near
| Pers3_Distant;
The verb inflects for the gender (masculine and feminine) and VTense parameter covers the
tenses of the Sindhi language. ‘Subj’, ‘Perf’ and ‘ Imperf’ forms describe the subjunctive ,
perfective and imperfective aspects. The present, past and future tenses are quoted here as
subjunctive, perfective and imperfective respectively. The verbs also inflect for the 1st
person , 2nd person (Casual and Respect) and 3rd person (Near and at Distance).
fun live _V : V ;
live_V = mkV " ھڻر " , rh ɳ ;
Here lexical paradigm is mkV, which takes lemma “ ھڻر , rhɳ” live_V and builds finite
inflection table by analyzing the lemma.
The different forms of the live_V are built by mkV, which are as follows:
s .VF Subj Pers1 Sg.Masc => ان ھر -- rha:n
s .VF.Imperf .Pers3_Distant .Sg .Masc => ندو ھر -- rhndo
..........
s .VF.Imperf .Pers3_Distant .Sg .Fem => ندي ھر --rhndi:
s . .Inf => ھڻر -- rhɳ
s . .Inf_Fem => ي ھر -- rhi:
s . .Inf_Obl => ھر -- rh
s . .Ablative => ون ھر -- rhon
s . .Root => ھر -- rh
9
2.1.3 Adjectives
Sindhi Adjectives inflect for gender, case and number. Sindhi adjectives are agreeing with the
noun with respect to the gender, case and number. The inflectional categories of Sindhi
adjectives are declinable and indeclinable. The declinable category adjectives inflect for the
case and gender of the noun, where as the indeclinable category of adjectives remain constant.
The adjectives can be substituted as an adverb.
Adjective = {s : Number => Gender => Case => Str} ;
‘s’ stores inflected forms of the adjective which inflect for the number (Singular, Plural), case
(Direct, Oblique, Vocative, Ablative) and gender (Masculine, Feminine).
fun wide_A : A
lin wide_A = mkA “رو ڪوی ” -- oi:kro
Here mkA is a lexical paradigm defined as:
mkAdj : Str -> Adjective ;
Here mkAdj accepts a string and stores after analyzing that string. It calls another function adj
which develops an inflectional table and stores all inflectional forms of an adjective. In Sindhi
each adjective has sixteen inflectional forms.
adj :(x1,_,_,_,_,_,_, _,_,_,_,_,_,_,_, x16 : Str) -> {s : Number => Gender => Case => Str} =
\msd,mso,msv,msa, fsd,fso,fsv,fsa, mpd,mpo,mpv,mpa, fpd,fpo,fpv,fpa -> {
s = table {
Sg => (cmnAdj msd mso msv msa fsd fso fsv fsa).s ;
Pl => (cmnAdj mpd mpo mpv mpa fpd fpo fpv fpa).s
}
};
cmnAdj : (x1,_,_,_,_,_,_,x8 : Str) -> {s : Gender => Case => Str} =
\sd,so,sv,sa, pd,po,pv,pa -> {
s = table {
Masc => table {
Dir => sd ;
10
Obl => so ;
Voc => sv ;
Abl => sa
} ;
Fem => table {
Dir => pd ;
Obl => po ;
Voc => pv ;
Abl => pa
}
}
} ;
The following table is showing different forms of adjective wide.
s . Sg => Masc => Dir => رو ڪوی -- oi:kro
s . Sg => Masc => Obl => ري ڪوی -- oi:kri:
s . Sg => Masc => Voc => را ڪوی -- oi:kra:
s . Sg => Masc => Abl => را ڪوی --oi:kra:
.....
s . Pl => Fem => Dir => ریون ڪوی --oi:kri:on
s . Pl => Fem => Obl => رین ڪوی -- oi:kri:n
2.1.4 Numerals
Sindhi numerals are written from left to right where as script is right to left. These numerals
possess a verity of numerals including Cardinal numbers and ordinal numbers. The cardinal
numbers describe the quantity and are used for counting in Sindhi where as ordinal numbers
refers to the ranking and ordering. For example:
Cardinal Number:
لي ٻ hk_one bli:_cat ,ھڪ
تاب ڪچار , ca:r_four kta:b_book
Ordinal Number:
تا آهن ي ڪمون و ٽ ٽ , mon_I ot ̪ _have ti:̪ _three kta:_dogs
الهائيندي آهي وليون ڳهو ست ٻ , ho_she st_seven boli:on_languages a:lha:ɡʱ ʔi:ndi: _speak
11
Cardinal Number Sindhi Ordinal Number Sindhi
One ڪه , hk First phri:on, پهریون
Two ɓ ,ٻـ Second ɓ i:on , ٻيون
Three ̪:ti ,ٽي Third ti:on̪ , ٽيون
Four ca:r ,چار Forth ون ٿچو , cot onʰ
Five pnɟ ,پنج Fifth pn , پنجون onɟ
Table 2.2: Sindhi Numerals
2.1.5 Adverbs
The Sindhi has some original adverbs. They never inflect and their structure never changes.
Sindhi adverbs are classified into five categories. These are temporal (indicate time of action),
locational (indicates place of action), directional (indicates direction of action), manner
(indicates manner of action) and degree (indicates degree of action) adverbs.
Adverb Categories Adverb
Temporal ي ,kdhn_when ,ڪڏھن ڻا :ha ,ھ i:_ɳ Now, ڄا , a: _ʄ today
Location و ,kti:_ Where ,ڪٿي ,hti:_Here,ھتي ھویج ,oi: o_ɟʱ Near, پري, pri:_ Away
Directional ن ان ھي ڏ ki:da:nhn_Where or in which direction,ڪ
Manner a:ei:n_In this manner, like this ,ائين
Degree و ھڻگ , o_ɡ ɳʱ Much a lot, یان ڍا :da , ڏ i:n_ɖʱ Loudly , ھوریانhori:a:n_Slowly
Table 2.3: Categories of Adverbs
GF uses following simple structure to store adverbs.
cat Adv ;
lincat Adv= { S : Str } ;
These are constructed by simple functions. For Example
fun today_Adv : Adv ;
today_Adv = mkAdv "aJ' " ;
mkAdv : Str -> Adv = \ str -> { s = str } ;
Here the function mkAdv gets input string and stores it in the field ‘s’.
12
2.1.6 Pronouns
Pronouns are substitute of the nouns. These inflect for case and number. Few pronouns mark
for the gender. Pronoun for person III is gender distinction.
Type Case/Masc Case /Fem
Sg Pl Sg Pl
Dir Obl Dir Obl Dir Obl Dir Obl
Personal Person I (I)
:ma,مـــان
n
mon,مون
,
ن +جو ھم
mhn+ oɟ
(We)
اسين
a:si:n
اسان
a:sa:n
(I)
مان
,ma:n
mon,مون
,
ن +جو ھم
mhn+ oɟ
(We)
اسين
a:si:n
اسان
a:sa:n
Personal Person
II
(You)
ton,تون
ن + جو ھت
thn+ oɟ
(You
)
ين ھتو
tohi:
n
ان ھتو
toha:
n
(You)
تون
,ton
(Yo)
تون
,ton
(You
)
تون
,ton
ان ھتو ,
toha:n
Table 2.4: Pronoun for 1st person and 2nd person
Types of pronoun are personal, reflexive, relative, interrogative and demonstrative.
Type
Personlll
Masc Fem
Sg Pl Sg Pl
Dir Obl Dir Obl Dir Obl Dir Obl
Demonstrative
(he)
ho,ھو
(this)
:hi,ھي
و ھا ,
a:o
,ھي
hi:
(they
these
)
ي ھا ,a
:i:
(they
)
ن ھان ,a:n
hn
ين ھان
ho ,ھو
'she
this'
:hi ,ھي
'she'
that
ا ھا
a:ha:
ho,ھو
they
these'
ي ھا ,a:hi
:
'they'
those'
ي ھان ,a:n
hi:
ـــــــي يئ ھا ,
a:hi:ei:
13
Reflective ڻپا ,pa:
ɳ
self
ن + ھپ
جو
phn+
oɟ
ڻپا
,
pa:
ɳ
ن + ھپ
جا
phn+ a:ɟ
ڻپا
pa:ɳ
ن + جي ھپ
phn+ i:ɟ
ڻپا
,pa:ɳ
ن + جا ھپ
phn+ a:ɟ
Interrogative ڪير
,ki:r,
who
ڇا
.
c a:,ʰ wh
at
ن ھن ڪ
,
knhn,
who
m
ڇا
.c a:,ʰ
what
ڪير
,ki:r,
who
ڇا
what
ڪيئن
ڇا
.
c a:,ʰ wh
at
ڪير
,ki:r,
who
ڇا
.
c a:,ʰ wh
at
ن ھن knh,ڪ
n, whom
ڇا
.
c a:,ʰ what
ڪير
,ki:r,
who
ڇا
.
c a:,ʰ wh
at
ڪيئن
,knhn,
whom
ڇا
.
c a:,ʰ wh
at
Relative ,جو
o,ɟ
who/
which
ن ھجن
nhnɟ
جي
,
i:ɟ ,
who/
whic
h
ي ڪجي
i:ki:ɟ ,
that
, جا
a:ɟ
'who'
which
ي ڪجي i:kɟ
i: , that
, جي
i:ɟ
who
ي ڪجي ,
i:ki:ɟ ,
that
Table 2.5: pronoun for 3rd Person
In GF pronoun inflects for case oblique and direct. The pronoun ‘we’ is defined in GF as
we_Pron = personalPN " اسان" , a:sa:n " اسين " , a:si:n ;
mkPron : (x1,x2 :Str) -> {s:Case => Str} =
\y1,y2 -> {
s = table { Dir => y1; Obl => y2; }
};
2.1.7 Conjunctions
The single phrase , words of sentence , two or more sentences stand to one another, to
express their relation conjunctions are used. These belongs to an un-inflected classes.
Conjunctions are divided at syntactic level in to two main categories, coordinate
conjunctions and subordinate conjunctions.
2.1.7.1 Coordinate conjunctions
14
They joini two mutually independent sentences. The core coordinate conjunctions are:
Copulative, Alternative and Adversative.
Coordinate conjunctions Examples
Copulative Also ڻِپ (pɳ)
Alternative Or یا (i:a:) , Neither و ڪن (nko)
Adversative But ي Either , (pr) پر ڙتو (to i:ɽ )
Table 2.6: Coordinate Conjunctions
2.1.7.2 Subordinate conjunctions
They are usd to construct a complex sentence by joining subordinate clauses. Some of
the core Subordinate conjunctions are time, direction, location, reason, condition,
manner etc.
Subordinate
conjunctions
Examples
Time When,,ن ڏھجـــــ ,( dhnɟ ),Before,ریـــــن ھپ ,(phri:n),Till,تـــــائين
(ta:ei:n)
Location Where ) جتي ti:ɟ )
Direction Where ن ان ھجي ڏ ( i:da:nhnɟ )
Manner As جيئن ( i:enɟ )
Reason There fore ري ڪان (a:nkri:), Because ڇو ت. (c o tʰ )
Table 2.7: Subordinate Conjunctions
2.1.8 Interjection
A word or a phrase that describes emotions or sudden feelings is an interjection. It is
always used in the vocative form. It does not have any relation with any other word in
a sentence grammatically. Basically it is a sentence itself. It expresses surprise, joy
and grief etc. Some interjections are as below.
Joy واہ واہ , oa:h oa:h_ Wao
Grief افسوس , a:fsos_sad
Surprise مار , ma:r_oo
Chapter3. Syntax
15
The morphology section describes different Sindhi lexical words and their different
inflectional forms. in this section words combination and making words will be discussed.
Sindhi language syntax is divided into two categories, analytical and synthetic.
3.1 Analytical Syntax
Analytical part explains how different parts of the speech join together according to their
exact meaning, essential value and particular purpose. It includes noun phrase, verb phrase,
adjective phrase and another close categories. The word order in neutral sentences in Sindhi is
subject-object-verb (SOV) but, word order of sentence can be changed like SOV, SVO, OSV,
VSO, etc. For example as shown bellow
ي لي ري ڏٺو ٻ ڪ c ,ڇ okri:_boy li:_catɓʰ dt i:_seen ̪ʰ SOV
لي ي ري ٻو ڏٺ ڪ c ,ڇ okri:_boy dt i:_seen li:_cat ɓʰ ̪ʰ SVO
ي ري و ڏٺلي ڪ ڇ . ٻ li:_cat c okri:_boyɓ ʰ dt i:_seen̪ʰ OSV
3.1.1 Noun Phrase (NP)
Sindhi noun phrase can be a word or it can be a bunch of words which are related to each
other grammatically and work together as a noun phrase. It has a head noun and complements
or modifiers. Noun phrases can be interpreted without determiner and its structure contains
modifiers and complements as Determiner-Modifier-Noun (تاب نئي ڪي hi:_This ,ھ nei:_new
kta:b_book) , Possesor-Quantifies-Noun ( تاب نجا چار ڪم ھ , mhn a:_ɟ my ca:r_ four kta:b_ book )
[10] .
Some noun phrases have an exact meaning on their own, like the Sun “ سج , sɟ”. All proper
nouns are noun phrases [11].
Here a noun phrase structure is described in GF. The noun phrase and its all forms are
recorded into two fields ‘s’ and ‘ a’ .
Lincat NP : Type = { S : NPCase => Str ; a : Agr ; };
Here ‘s’ is the inflectional table storing the record from NPCase to string (NPCase => Str)
and ‘a ’ is a feature of the noun phrase and it is used to select the appropriate form of other
categories which matches with the noun.
Param NPCase = NPC Case | NPNom | NPVoc | NPIns1 | NPIns2
| NPGen | NPDat | NPAcc | NPLoc1 | NPLoc2 | NPAbl ;
param Case = Dir | Obl | Voc | Abl ;
16
Param Agr = Ag Gender Number Pperson ;
Param Gender = Masc | Fem;
Param Pperson = Pers1| Pers2_Casual | Pers2_Respect
| Pers3_Near | Pers3_Distant;
Param Number = Sg | Pl
Here the constructor ’NPC Case’ stores the cases of noun phrase.
NPC Case : It returns the cases of noun as in lexicon.
NPNom : Nominate case with the case creator ‘ i:ɟ oɟ , جي .‘ جو
NPVoc : Vocative case is used to address an inferior person with case creator ‘a:o : او ‘ .
NPIns1 : Instrumental case is used to define an action performed by noun with case creator ‘
ka:n : کان ‘ .
NPIns2 : Instrumental case is used to define noun accompanied with any thing with case
creator ‘sa:n : سان ‘ .
NPGen : Genitive case with suffix case creator ‘ i:ɟ oɟ , جي . ‘ جو , sa:n سان
NPDat : Dative Case is used to denote the postposition case creator ‘ ki: کي’ .
NPAcc : Accusative case is used as nominative case or dative case with postposition case
creator ‘ ki: کي’.
NPLoc1 : Locative case1 is used to define place and time with case creator ‘ ti: تي ‘.
NPLoc2 : Locative case2 is used to define place with case creator ‘ mi: : ہ ‘
NPAbl : Ablative case is used to define place with post position case creator ‘ta:n تان, ma:n
. ’مان
Postpositions are placed after nouns to show their relationship with other constituents of the
sentence. The lexical category of noun is changed first into an intermediate category common
noun. Then common noun is converted into a noun phrase. A common noun ‘CN’ deals with
noun and its modifiers. For example :
fun AdjCN : AP -> CN -> CN ;
Lin AdjCN ap cn = {
s = \\n,c => ap.s ! n ! cn.g ! c ++ cn.s ! n ! c ;
g = cn.g
} ;
The AdjCN linearization results a common noun for example ‘grm_Hot ca:nh_Tea چان ھگرم
garam chai ’. Here ‘ ca:nh ھچان ’ is common noun which is modified by the Adjective Phrase
(AP) ‘ گرم ,grm’ .
17
As Sindhi adjective inflects for number, gender, case and degree so that the appropriate form
of the adjective concatenate which agrees with the common noun. It is ensured by choosing a
particular form of noun and an adjective from inflecting table, by using the selection operator
(!). The Noun phrase can be obtained by converting common noun by using different
functions.
Some functions which can be used to construct the noun phrase are as bellow:
fun DetCN : Det -> CN -> NP ; -- the Woman
fun UsePN : PN -> NP ; -- John
fun UsePron : Pron -> NP ; -- She
fun MassNP : CN -> NP ; -- (water)
The noun phrase can be built in different ways that are so common in other languages, as in
the abstract syntax resource grammar but the linearization is defined in the concrete syntax
because it is the language dependent.
3.1.2 Verb Phrase (VP)
Sindhi verb phrase is a syntactic form. It has a complex structure. The main verb is complex
followed by an auxiliary verb which marks mood and tense (as object of the verb, adverb etc),
these are used in development of other categories and clauses. A verb phrase structure is
described in GF as:
VPH : Type = {
s : VPHForm => {fin, inf : Str} ;
subj : VType ;
obj : {s : Str ; a : Agr} ;
comp : Agr => Str;
ad : Str;
inf : Str;
embComp : Str ;
prog : Bool ;
} ;
The Verb phrase is recorded in the inflectional table ‘s’ having two strings {fin, inf : Str}
storing all the values of Parameter VPHForm that is described as follows:
18
Param VPHForm = VPTense VPPTense Agr| VPReq| VPImp | VPReqFut | VPInf |
VPStem ;
param VPPTense = VPFutr |VPPast | VPPres |VPPerf;
The VPHForm has six cases. The first constructor VPTense shows all the inflectional forms
which are inflecting for the number, person, gender and tense (VPPTense) with the second
constructor agreement feature (Agr). The third constructor (VPReq) shows a request form of
the verb phrase.
The forth (VPImp) shows an imperative form of the VP. The fifth constructor (VPInf)
contains the infinitive form. The sixth constructor VPStem contains a root form of the verb.
These all constructors do not occur at the same time in Sindhi language.
The ‘inf ‘ stores an appropriate required form of the verb which agrees with the tense and the
‘fin ‘ stores the copula or auxiliary verb. The ‘obj’ stores an object of the verb and its
information. The ‘subj’ stores transitivity of the verb by means of VType (Intransitive,
Transitive or transitivePost):
Param VType = VIntrans | VTrans | VTransPost ;
The ’Comp’ stores the complement of verb which also inflects for the gender, number and
person with Agr. The ‘ad’ stores the adverb.
The ‘embCom’ deals with exceptions of the word order in Sindhi language during the
development of clauses and stores embedded complements. For example, if a sentence is a
complement of a verb then it takes a different place in a clause; that is it places at the end of
the clause.
و ان ي ت مان راند کي ٿن چيو ا ڏ ھ ھ
hn_ he ci:o _ says a:hi:_ Aux t_ that ma:n_ I ra:nd_ game kʰi:da:n_ play to_ Aux
He says that I play game.
If an adverb is used as a complement of the verb then it is placed before the main verb.
ي ندي ا و و تيز ي ت ھن چيو ا ڙ ڊ ھ ھ ھ
hn_ she ci:o _ says a:hi:_ Aux t_ that
ho_ she ti:z_ fast ɖorndi:_ run ahi:_ Aux
She says that she runs fast.
Through the development of different functions, the lexical form of the verb is changed to the
syntactic form (VP).
fun UseV : V → VP ;
lin UseV v= PredV v ;
19
PredV is a function which converts the lexical form (V) of verb to the syntactic form of verb
phrase.
predV : Verb -> VPH = \verb -> {
s = \\vh => case vh of {
VPTense VPPres (Ag g n p) => {fin = copula CPresent n p g ; inf = verb.s ! VF Subj p n g } ;
VPTense VPPast (Ag g n p) => {fin = copula CPast n p g ; inf =verb.s ! VF Perf p n g} ;
VPTense VPFutr (Ag g n p) => {fin = copula CFuture n p g ; inf = verb.s ! VF Imperf p n
g } ;
VPTense VPPerf (Ag g n p) => { fin = [] ; inf = verb.s ! Root ++ cka g n } ;
VPStem => {fin = [] ; inf = verb.s ! Root};
VPInf => {fin = verb.s!Inf_Obl ; inf = verb.s ! Root};
VPImp => {fin = verb.s!VF Subj Pers3_Near Pl Masc ; inf = verb.s ! Root};
VPReq => {fin = [] ; inf = verb.s!VF Subj Pers1 Pl Masc};
_ => {fin = [] ; inf = verb.s ! Root}
};
obj = {s = [] ; a = defaultAgr} ;
subj = VIntrans ;
inf = verb.s ! Inf;
ad = [];
embComp = [];
prog = False ;
comp = \\_ => []
} ;
The verb has three lexical forms. These forms are used to make tenses (VPPres, VCPPast,
VPFutr, VPPer ). All these forms cover different combinations of the tenses, mood and aspect
of Sindhi.
The above code builds an inflectional table ‘s’ for the present tense at verb phrase level
‘VPPres’ of all possible forms of the number, gender and person(Ag p g n ). In the above
code, the lexical imperfective form (VF Imperf n g p) of verb (V) is used to make
present tense. As described above ‘inf’ stores the main verb form and ‘fin’ stores the
corresponding auxiliary verb. In the above code all other parts initialize with empty or default
values. These parts improve the verb phrase with another component e.g. an adverb
complement etc.
20
Want to eat, khaen: _eat chahen:_ want
Compl VV : VV → VP → VP ;
Sit here , hity_ here Weh:_ sit
AdvVP : VP → Adv → VP
3.1.3 Adjectival Phrase (AP)
The Sindhi adjectives inflect for the number case and gender at morphological level. At the
syntax level adjectival phrase agree with the noun. In GF adjectival phrases are described as
bellow.
fun PositA : A -> AP ; -- (cold, thado)
The linearization of the adjectival phrase is a very simple and resembles with an adjective.
lin positA a = a ;
The other way to develop ‘AP’ is as following:
fun ComparA : A -> NP -> AP ;
A comparative adjectival phrase is developed from a noun phrase and an adjective. The
linearization of the above function in GF is as follows:
ComparA a np = {
s = \\n,g,c => np.s ! NPC Obl ++ "k'an" ++ a.s ! n ! g ! c ;
s = \\n,g,c => np.s ! NPC Abl ++ "k'an" ++ a.s ! n ! g ! c ;
} ;
Other examples of adjectival phrase functions are as shown below:
As sweet as I , etro_ As mitho_ sweet jetro_ As man_ I
fun CAdvAP : CAdv -> AP -> NP -> AP ; -- as cool as John
Older then I , mun_ I kan_ then porho_ old
fun ComparA : A -> NP -> AP ;
3.1.4 Adverbs
Sindhi adverbs never inflect so their construction is very simple. At syntactical and lexical
level there are different close classes and function constructions. For example, the
construction of an adverb is as bellow:
PositAdvAdj : A -> Adv ;
Warmly , garam:
21
AdAdv : AdA -> Adv -> Adv ;
o_ɡ ɳʱ Very ti:zi:_quickly (و تيز ھڻگ )
3.1.5 Copula (Auxiliary verb)
A copula modifies the main verb in a clause to form a particular tense. It defines a link
between the subject of a sentence and a predicate (an adverbial or a subject complement).
Sindhi copulas inflect for the tense, number, gender and person.
In GF auxiliary verb is implemented for subjective, perfective, imperfective and continuous
tense.
param CTense = CPresent | CPast | CFuture | CContinuous;
oper
copula : CTense -> Number -> PPerson -> Gender -> Str = \t,n,p,g ->
case <t,n,p,g> of {
<CPresent,Sg,Pers1,Masc > => "ٿو "; -- t oʰ
<CPresent,Sg,Pers1,Fem > => " ٿي " ; -- t i:ʰ
<CPresent,Sg,Pers2_Casual,Masc > => "ٿو " ; -- t oʰ
<CPast,Pl,Pers2_Casual,_ > => "ھئا " ; -- hea:
<CPast,Pl,Pers2_Respect,_ > => "ھيو " ; -- hi:o
..........
<CPast,Pl,Pers3_Near,Masc > => "ھئا " ; -- hea:
<CPast,Pl,Pers3_Near,Fem > => " ھئيون " ; -- hei:on
<CPast,Pl,Pers3_Distant,Masc > => "ھئا " ; -- hea:
<CPast,Pl,Pers3_Distant,Fem > => "ھيون " ; -- hi:on
} ;
Here (t,n,p,g) are the veriables which store different inflectional forms of the copula.
Here ‘t’ describes the tense, ‘n’ defines the number which is either singular or plural,
PPerson defines the 1st eprson or 2nd person or 3rd person, and ‘g’ is used to define
gender of the copula.
In GF, eighty inflectional forms of the copulas are implemented. Those copulas are
very useful to make meaninGFul sentences.
22
3.2 Synthetical Syntax
The synthetical syntax describes that how parts of the speech join together to form a single
clause or sentence, two clause or sentences or more than two. [10]
Sindhi Phrases are head-final. Generally the verb, post position and noun with complements
and modifiers precede their heads.
ایان ي مو ٽمان ماء پئي جو چو ن ٿ ڻ
I do not reject the word of mother and father.
ma:n ma: pi: o co n t i: mot a:i:a:nʔ ʔ ɟ ɳ ʰ ̪
3.2.1 Clauses
Clauses are included in the synthetical part of the syntax. As Sindhi phrase is a lexical word
or combination of the lexical words and these words are related to each other Grammatically.
A clause is single phrase or combination of the phrases which are linked together. Clause can
be formed by joining the different types of the phrases including noun phrase, verb phrase,
adjective phrase etc. A Clause may not have any subject sometimes or it may have a predicate
and a subject of its own. These clauses are used to form a sentence. There is a difference
between a sentence and a clause in the Grammatical Framework resource grammar API
tense system. That is a clause has inconsistent tense whereas a sentence have fixed tense.
For the construction of a sentence in GF , clauses are developed first and then clause tenses
which are used to make sentences.
lincat Clause : Type = {s : VPHTense => Polarity => Order => Str} ;
Where :
Param VPHTense = VPGenPres | VPImpPast | VPFut
| VPContPres | VPContPast | VPContFut
| VPPerfPres | VPPerfPast | VPPerfFut
| VPPerfPresCont | VPPerfPastCont | VPPerfFutCont
| VPSubj ;
Param Order = ODir | OQuest ;
Param Polarity = POS | Neg ;
23
fun PredVP : NP -> VP -> Cl ; -- Rana walks
Its linearization is :
lin predVP np vp = mlClause np vp ;
The GF resource library does not cover whole tense system of the Sindhi, however that is
structured to cover more or less the aspect and mood of the tense system. It covers only eight
combinations along with present tense, past tense, future tense and conditional.
To achieve more and more coverage of the Sindhi tense system, twelve tense (VPHTense) are
developed based on sentences at clause level. The parameter Order is used to develop
questions and direct clauses whereas parameter polarity is used to develop negative and
positive clauses. The SOV agreement of the analytical syntax is accessible by using predVP.
The SOV agreement is a feature of the Sindhi and its behaviour is ergative which is covered
in noun phrase (NP).
where
oper mkClause : NP -> VPH -> Clause = \np,vp -> { s = \\vt,b,ord =>
Subj agr : NPCase * Agr = case vt of {
VPImpPast => case vp.subj of {
VTrans => <NPErg, vp.obj.a> ;
VTransPost => <NPErg, defaultAgr>;
_ => <NPC Dir, np.a>
} ;
_ => <NPC Dir, np.a>
} ;
quest =
case ord of { ODir => []; OQuest => "c'a" };
in
quest ++ np.s ! subj ++ vp.obj.s ++ vp.ad ++ vp.comp ! np.a ++ vps.inf ++ n++ vps.fin ++
vp.embComp
} ;
The above code takes a noun and verb phrase and then construct a clause having a polarity,
order and tense and that is fixed in sentences by using different functions. The noun phrase
selects an appropriate form of the verb phrase.
24
fun UseCl : Temp → pol → Cl → S ;
Here, the syntactic category is Temp, it is used to store a record of tense and anteriority. In
Temp category, the tense refers to the abstract level tense and it is used for Sindhi tense by
selecting the particular clause.
3.2.2 Sentences
Sentences cover an important and main part of synthetical syntax. Sindhi sentences are
divided in two major categories: simple sentence and compound sentence.
Simple sentences must contain a subject and a predicate. The Subject may be defined by
adjective or subjective or numeral or pronoun. The predicate may be defined by an adjective
or verb or subjective or numeral with the copula.
The compound sentence is formed when two or more than two sentences are joined together.
It is done either by the way of subordination or coordination.
Coordination sentences are formed by joining two or more sentences and each sentence
remains an independent of the other. It is formed without or by the copulative, disjunction and
by the negative adverb (Neither-Nor نـ ).
The Subordinate sentences are formed when two or more sentences may be joined in such a
way that one of the sentence depends on an other sentence where as the other sentence should
be an independent.
As a clause have an inconsistent tense, order and polarity so through different functions the
parameters like order and polarity can be fixed which results into declarative sentence.
In GF, generation of sentence construction is as follow:
UseCl temp p cl =
{ s = case <temp.t,temp.a> of {
<Pres,Simul> => temp.s ++ p.s ++ cl.s ! VPGenPres ! p.p ! ODir;
<Pres,Anter> => temp.s ++ p.s ++ cl.s ! VPPerfPres ! p.p ! ODir;
<Past,Simul> => temp.s ++ p.s ++ cl.s ! VPImpPast ! p.p ! ODir;
<Past,Anter> => temp.s ++ p.s ++ cl.s ! VPPerfPast ! p.p ! ODir;
<Fut,Simul> => temp.s ++ p.s ++ cl.s ! VPFut ! p.p ! ODir;
<Fut,Anter> => temp.s ++ p.s ++ cl.s ! VPPerfFut ! p.p ! ODir;
<Cond,Simul> => temp.s ++ p.s ++ cl.s ! VPSubj ! p.p ! ODir;
<Cond,Anter> => temp.s ++ p.s ++ cl.s ! VPSubj ! p.p ! ODir;
};
} ;
25
The following function is one of the most important functions:
fun UseCl : Temp -> Pol -> Cl -> S ;
Here ‘Temp’ parameter is combination of the parameters for tense and for interiority. The
function ‘UseCl’ accepts input as polarity, anteriority, tense and a clause and it gives output as
a sentence. The different types of the sentences those are relative sentences and interrogative
sentences are developed through the function as follows:
fun UseRCl : Temp -> Pol -> RCl -> RS ;
fun UseQCl : Temp -> Pol -> QCl -> QS ;
Chapter4. Applications
The reported resource grammar can be used as a resource for language processing
applications as natural language interfaces, translation system, multilingual generations and
spoken dialogue systems.
The minibar online is web application that allows to translate text into impelemented all
languages. There are menu for selecting the output and the input languge to translate. The
abstract shows the sentence parse tree and its linearization in Sindhi . Some snapshots of
Sindhi to English and English to Sindhi translations are shown bellow.
26
Chapter5.Evaluations
Testing a resource grammar is different from testing natural lanuge processing applications in
general, where testing is done against some text corpus. Testing resource grammars is much
like testing software libraries in general (Ranata 2009b). For checking the accuracy, the
Sindhi resource garammar used to observe and translate. The accuracy of Sindhi resource
grammar 97% acieved, because implemented resource grammar does not cover all the aspects
of the language.
The noun phrase cases formed by noun and ergative cases are not yet implemented fully. It
might be possible that reported grammar could not translate correctly.
Chapter6. Conclusion
Grammatical Framework resource grammar for Sindhi is the only translation platform
which provides the Grammatical translation results. Before this implementation, there was
not any particular tool that could provide the facility for text translation into Sindhi language.
Sindhi grammar library present in Grammatical Framework has used different categories and
functions to manage the morphology and syntax implementation. The library has 44
categories and 190 functions. This grammar implementation covers a major part of the Sindhi
language. As GF has an abstract syntax that is supporting a group of different languages, so it
is very difficult to cover each and every aspect of the language in the library of resource
grammar. The grammar of Sindhi language does not cover all aspects of language because
limited lexicon word list is not enough to cover the whole language. There is a need to
implement full Sindhi grammar and its all aspects that can translate the text more accurately.
There is also a need to extend the resource grammar library boundaries so that full grammar
of any language can be implemented .It is possible to increase the efficiency and effectiveness
of the Sindhi grammar by implementations of extra features of language.
30
Chapter7. Future Work
As mentioned already this resource grammar does not cover all the features of the Sindhi. In
the future these uncovered features and aspects can be implemented and developed. As
lexicon is not a full vocabulary of Sindhi language, one can increase lexicon vocabulary list as
well.
The grammar resource can be extended by implementing a specific module for specific
features of Sindhi language. It can not be accessed through the common API of GF, but it can
be accessed in a particular language application grammar.
In general, the assessment of the resource grammar is an assessment of the software library. It
is very different from the natural language processing assessment, where the assessment is
generally done on some text croups. It is possible to develop a user friendly interface that can
be easily accessed by native speakers and they can take benefit from this development.
31
References
1. http :// www . britannica . com / EBchecked / topic /545670/ Sindhi -
language /282506/ Linguistic - features
2. http :// linguistics . ucsc . edu / about / what - is - linguistics . html
3. Ranta, A.: 2004c, ‘Grammatical Framework Tutorial’. In: A. Beckmann and
N.Preining (eds.): ESSLLI 2003 Course Material I, Vol. V of Collegium Logicum.pp.
1–86, Kurt G¨odel Society.
4. http :// www . Grammatical framework . org /
5. http://en.wikipedia.org/wiki/Grammatical _Framework
6. Ranta. 2009a. Grammatical Framework: A Multilingual Grammar Formalism,
Language and Linguistics Compass, Vol. 3.
7. A. Ranta. 2009b. Grammars as Software Libraries.In Y. Bertot, G. Huet, J-J.
Lévy, and G. Plotkin (eds.), From Semantics to Computer Science,Cambridge
University Press, pp. 281-308.
8. Memon, G.Q.; Shah, S.H.A.; Dept. of Comput. Sci., Shah Abdul Latif Univ.,
Khairpur, Pakistan (May 2010. Algorithms for Sindhi Word Segmentation using
Lexicon-Driven Approach, International journal of Academic Research , Volume 3.
Print ISBN: 978-0-7695-4043-6 .
9. Rahman, M. U (2009). Sindhi Morphology and Noun Inflections, conference on
language & technology, center for research in Urdu language processing, Lahore,
Pakistan: 74-81.
10. C. Jennier. (2006), The Sindhi Language. In K. Brown (ed). Encyclopedia of language
and Linguistics, 2nd Edition, v. 11:384-386. Oxford: Elsevier.
11. Trumpp, Ernest. 1872. Grammar of the Sindhi Language. London: Tru<bner and Co.
and Leipzig: FA Brockhaus.
12. International Journal of Computer Theory and Engineering, Vol. 2, No. 4, August,
2010 1793-8201 Sindhi Part of Speech Tagging System Using Wordnet. Javed Ahmed
Mahar and Ghulam Qadir Memon .
13. A. Ranta (2009): Grammatical Framework: A Programming Language for
Multilingual Grammars and Their Applications.
14. Aarne(Dec 2009): The GF Resource grammar library, Linguistic Issues in Language
Technology – LiLT , Volume 2, issue 2.
32
15. Machine Translation Summit VII, 13th-17th September 1999, Kent Ridge Labs,
Singapore.Proceedings of MT Summit VII "MT in the great translation era" (Tokyo:
AAMT), pp.30-44.
16. Shafqat M. Virk, M. Humayoun, A. Ranta., 2010. "An Open Source Urdu Resource
Grammar". Proceedings of the 8th Workshop on Asian Language Resources. In
conjunction with Coling 2010.
17. Shafqat M. Virk, M. Humayoun, A. Ranta., 2011. "An Open Source Punjabi Resource
Grammar". Proceedings of RANLP 2011 (Recent Advances in Natural Language
Processing), pages 70-76, Bulgaria, 12-14 September 2011.
33
Appendix
Lang: PositA warm_A
LangEng: warm
LangSnd: گرم
Lang: ComparA warm_A (UsePron i_Pron)
LangEng: warmer than I
LangSnd: مان کان گرم
Lang: SentAP (PositA good_A) (EmbedS (UseCl (TTAnt TPres ASimul) PPos (PredVP
(UsePron she_Pron) (UseComp (CompAdv here_Adv)))))
LangEng: good that she is here
LangSnd: و و ت ھس ی۶ٺ تی ا ھ ھ
Lang: AdAP very_AdA (PositA warm_A)
LangEng: very warm
LangSnd: تمام گرم
Lang: PositAdvAdj warm_A
LangEng: warmly
LangSnd: گرم
Lang: PrepNP in_Prep (DetCN (DetQuant DefArt NumSg) (UseN house_N))
LangEng: in the house
LangSnd: ر ہ ھگ
Lang: ComparAdvAdj more_CAdv warm_A (UsePN john_PN)
LangEng: more warmly than John
LangSnd: ی گرم ا و ڪجان گ ڌ ھڻ
Lang: ComparAdvAdjS more_CAdv warm_A (UseCl (TTAnt TPres ASimul) PPos (PredVP
(UsePron he_Pron) (UseV run_V)))
LangEng: more warmly than he runs
LangSnd: و ی و و ی گرم ا و ٿگ ڙ ڊ ھ ڪ ڌ ھڻ
34
Lang: SubjS when_Subj (UseCl (TTAnt TPres ASimul) PPos (PredVP (UsePron she_Pron)
(UseV sleep_V)))
LangEng: when she sleeps
LangSnd: و ن ھج ی۶ڏھ ی ٿ سم ھ
Lang: AdNum (AdnCAdv more_CAdv) (NumNumeral (num (pot2as3 (pot1as2 (pot0as1
(pot0 n5))))))
LangEng: more than five
LangSnd: ی پنج ڪسان گ و ڌ ڏ
Lang: ConjS and_Conj (BaseS (UseCl (TTAnt TPres ASimul) PPos (PredVP (UsePron
he_Pron) (UseV walk_V))) (UseCl (TTAnt TPres ASimul) PPos (PredVP (UsePron
she_Pron) (UseV run_V))))
LangEng: he walks and she runs
LangSnd: و و ہ لی ھھو ٿھھھ ھ ی۶ھ ی و ٿ ڙ ڊ
Lang: ConjAP and_Conj (BaseAP (PositA cold_A) (PositA warm_A))
LangEng: cold and warm
LangSnd: ٿڌھھھو ہ گرم
Lang: ConjNP or_Conj (BaseNP (UsePron she_Pron) (UsePron we_Pron))
LangEng: she or we
LangSnd: یا اسان۶ھو
Lang: ConjAdv or_Conj (BaseAdv here_Adv there_Adv)
LangEng: here or there
LangSnd: ھتی یا اتی
Lang: ConjS either7or_DConj (BaseS (UseCl (TTAnt TPres ASimul) PPos (PredVP
(UsePron he_Pron) (UseV walk_V))) (UseCl (TTAnt TPres ASimul) PPos (PredVP (UsePron
she_Pron) (UseV run_V))))
LangEng: either he walks or she runs
LangSnd: و۶ڪو و یا لی و يو ھی ٿ ھ ھ ی۶ٻ ی و ٿ ڙ ڊ
Lang: ConjAP both7and_DConj (BaseAP (PositA warm_A) (PositA cold_A))
35
LangEng: both warm and cold
LangSnd: و۶ٻ ٿڌی گرم ہ
Lang: ConjNP either7or_DConj (BaseNP (UsePron he_Pron) (UsePron she_Pron))
LangEng: either he or she
LangSnd: و۶ڪو و یا يو ھی ھ ۶ٻ
Lang: ConjAdv both7and_DConj (BaseAdv here_Adv there_Adv)
LangEng: both here and there
LangSnd: تی ہ اتی۶ٻ ھی
Lang: ImpersCl (UseComp (CompAP (PositA hot_A)))
LangEng: it is hot
LangSnd: ی ھوسو ا ڪ
Lang: GenericCl (UseV sleep_V)
LangEng: one sleeps
LangSnd: و۶ڪو ی ٿی سم ھ
Lang: CleftNP (UsePron i_Pron) (UseRCl (TTAnt TPast ASimul) PPos (RelVP IdRP
(ComplSlash (SlashV2a do_V2) (UsePron it_Pron))))
LangEng: it is I who did it
LangSnd: يم رون ن ا يان ج ھمان ا ڪ ھ ھڙ ھ
Lang: CleftAdv here_Adv (UseCl (TTAnt TPast ASimul) PPos (PredVP (UsePron she_Pron)
(UseV sleep_V)))
LangEng: it is here that she slept
LangSnd: و ھتی يل ۶ھ ھ سم ی۶ھ
Lang: ExistNP (DetCN (DetQuant IndefArt NumSg) (UseN house_N))
LangEng: there is a house
LangSnd: ی ر ا ھتی گ ھ ھ
Lang: ExistIP (IdetCN (IdetQuant which_IQuant NumPl) (UseN house_N))
36
LangEng: which houses are there
LangSnd: ن ر ا ا گ ھتی جي ھ ڪ ھ
Lang: PredVP (UsePron i_Pron) (ProgrVP (UseV sleep_V))
LangEng: I am sleeping
LangSnd: و ان پيا ٿمان سم ھ
Lang: DetCN (DetQuant DefArt NumSg) (UseN man_N)
LangEng: the man
LangSnd: و ڻھما
Lang: UsePN john_PN
LangEng: John
LangSnd: جان
Lang: UsePron he_Pron
LangEng: he
LangSnd: ھو
Lang: PredetNP only_Predet (DetCN (DetQuant DefArt NumSg) (UseN man_N))
LangEng: only the man
LangSnd: و ڻھصرف ما
Lang: PPartNP (DetCN (DetQuant DefArt NumSg) (UseN man_N)) see_V2
LangEng: the man seen
LangSnd: و ڻھسيل ما ڏ
Lang: DetNP (DetQuant this_Quant (NumCard (NumNumeral (num (pot2as3 (pot1as2
(pot0as1 (pot0 n5))))))))
LangEng: these five
LangSnd: ھی پنج
Lang: DetCN (DetQuantOrd this_Quant (NumCard (NumNumeral (num (pot2as3 (pot1as2
(pot0as1 (pot0 n5))))))) (OrdSuperl good_A)) (UseN man_N)
37
LangEng: these five best men
LangSnd: و و ما ڻھی پنج س ٺ ھ
Lang: DetCN (DetQuant this_Quant (NumCard (NumNumeral (num (pot2as3 (pot1as2
(pot0as1 (pot0 n5)))))))) (UseN man_N)
LangEng: these five men
LangSnd: و ڻھی پنج ما ھ
Lang: DetCN (DetQuant this_Quant NumPl) (UseN man_N)
LangEng: these men
LangSnd: و ڻھی ما ھ
Lang: DetCN (DetQuant this_Quant NumSg) (UseN man_N)
LangEng: this man
LangSnd: و ھی ڻھما
Lang: NumCard (NumNumeral (num (pot2as3 (pot1as2 (pot0as1 (pot0 n5))))))
LangEng: five
LangSnd: پنج
Lang: NumCard (NumDigits (IIDig D_5 (IDig D_1)))
LangEng: 5 1
LangSnd: ۱ &+ ۵
Lang: NumCard (NumNumeral (num (pot2as3 (pot1as2 (pot1plus n5 pot01)))))
LangEng: fifty - one
LangSnd: ونجا ھای ڪ
Lang: NumCard (AdNum almost_AdN (NumDigits (IIDig D_5 (IDig D_1))))
LangEng: almost 5 1
LangSnd: ری و ڪگ ۵ &+ ۱ھڻ
Lang: OrdDigits (IIDig D_5 (IDig D_1))
LangEng: 5 1st
LangSnd: ۱ &+ ۵
38
Lang: OrdNumeral (num (pot2as3 (pot1as2 (pot1plus n5 pot01))))
LangEng: fifty - first
LangSnd: ونجا &+ ون ھای ڪ
Lang: OrdSuperl warm_A
LangEng: warmest
LangSnd: گرم
Lang: DetCN (DetQuantOrd DefArt (NumCard (NumNumeral (num (pot2as3 (pot1as2
(pot0as1 (pot0 n5))))))) (OrdSuperl good_A)) (UseN man_N)
LangEng: the five best men
LangSnd: و و ما ڻھپنج س ٺ
Lang: DetCN (DetQuant DefArt (NumCard (NumNumeral (num (pot2as3 (pot1as2 (pot0as1
(pot0 n5)))))))) (UseN man_N)
LangEng: the five men
LangSnd: و ڻھپنج ما
Lang: DetCN (DetQuant IndefArt (NumCard (NumNumeral (num (pot2as3 (pot1as2 (pot0as1
pot01))))))) (UseN man_N)
LangEng: one man
LangSnd: و ھڪ ڻھما
Lang: DetCN (DetQuant DefArt (NumCard (NumNumeral (num (pot2as3 (pot1as2 (pot0as1
pot01))))))) (UseN man_N)
LangEng: the one man
LangSnd: و ڻھ ما ھڪ
Lang: DetCN (DetQuant DefArt NumSg) (UseN man_N)
LangEng: the man
LangSnd: و ڻھما
Lang: DetCN (DetQuant DefArt NumPl) (UseN man_N)
LangEng: the men
39
LangSnd: و ڻھما
Lang: MassNP (UseN beer_N)
LangEng: beer
LangSnd: شراب
Lang: DetCN (DetQuant (PossPron i_Pron) NumSg) (UseN house_N)
LangEng: my house
LangSnd: ر ھگ
Lang: UseN house_N
LangEng: house
LangSnd: ر ھگ
Lang: AdjCN (PositA big_A) (UseN house_N)
LangEng: big house
LangSnd: ر و گ ھو ڏ
Lang: RelCN (UseN house_N) (UseRCl (TTAnt TPast ASimul) PPos (RelSlash IdRP
(SlashVP (UsePN john_PN) (SlashV2a buy_V2))))
LangEng: house which John bought
LangSnd: يو ریل ن جان ئرید ر جن ھگ ڪ ھ ھ
Lang: AdvCN (UseN house_N) (PrepNP on_Prep (DetCN (DetQuant DefArt NumSg) (UseN
hill_N)))
LangEng: house on the hill
LangSnd: ان ری م ر ٿگ ٽڪ ھ
Lang: SentCN (UseN question_N) (EmbedQS (UseQCl (TTAnt TPres ASimul) PPos
(QuestIAdv where_IAdv (PredVP (UsePron she_Pron) (UseV sleep_V)))))
LangEng: question where she sleeps
LangSnd: سواl و ھی ی۶ڪٿ ی ٿ سم ھ
40
Lang: DetCN (DetQuant (PossPron i_Pron) NumSg) (ApposCN (UseN friend_N) (UsePN
john_PN))
LangEng: my friend John
LangSnd: دوست جان
Lang: num (pot2as3 (pot1as2 (pot0as1 (pot0 n6))))
LangEng: six
LangSnd: ڇھ
Lang: num (pot2as3 (pot1as2 (pot0as1 pot01)))
LangEng: one
LangSnd: ھڪ
Lang: num (pot2as3 (pot1as2 (pot1 n6)))
LangEng: sixty
LangSnd: ٺھس
Lang: num (pot2as3 (pot1as2 pot110))
LangEng: ten
LangSnd: ڏھ
Lang: num (pot2as3 (pot1as2 pot111))
LangEng: eleven
LangSnd: ن ھیارن
Lang: num (pot2as3 (pot1as2 (pot1to19 n6)))
LangEng: sixteen
LangSnd: ن ھسورن
Lang: num (pot2as3 (pot1as2 (pot1 n6)))
LangEng: sixty
LangSnd: ٺھس
Lang: num (pot2as3 (pot1as2 (pot1plus n6 (pot0 n5))))
41
LangEng: sixty - five
LangSnd: ھٺپنج
Lang: num (pot2as3 (pot2 (pot0 n4)))
LangEng: four hundred
LangSnd: چار سو
Lang: num (pot2as3 (pot2plus (pot0 n4) (pot1plus n6 (pot0 n7))))
LangEng: four hundred and sixty - seven
LangSnd: ھٺچار سو ست
Lang: num (pot3 (pot2plus (pot0 n4) (pot1plus n6 (pot0 n7))))
LangEng: four hundred and sixty - seven thousand
LangSnd: چار l زار ھک ست ھٺ
Lang: IDig D_8
LangEng: 8
LangSnd: ۸
Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PPos (PredVP (UsePN
john_PN) (UseV walk_V)))) NoVoc
LangEng: John walks
LangSnd: و لی ٿجان ھ
Lang: UttQS (UseQCl (TTAnt TPres ASimul) PPos (QuestCl (PredVP (UsePron it_Pron)
(UseComp (CompAP (PositA good_A))))))
LangEng: is it good
LangSnd: ی و ا ا س ھا ا ٺ ھ ڇ
Lang: UttImpPol PNeg (ImpVP (UseV sleep_V))
LangEng: don't sleep
LangSnd: ھن سم
Lang: UttIP whoPl_IP
42
LangEng: who
LangSnd: ڪير
Lang: UttIP whoSg_IP
LangEng: who
LangSnd: ڪير
Lang: UttIAdv why_IAdv
LangEng: why
LangSnd: ڇو
Lang: UttNP (DetCN (DetQuant this_Quant NumSg) (UseN man_N))
LangEng: this man
LangSnd: و ھی ڻھما
Lang: UttAdv here_Adv
LangEng: here
LangSnd: ھتی
Lang: UttVP (UseV sleep_V)
LangEng: to sleep
LangSnd: ھڻسم
Lang: VocNP (DetCN (DetQuant (PossPron i_Pron) NumSg) (UseN friend_N))
LangEng: , my friend
LangSnd: دوست
Lang: QuestCl (PredVP (UsePN john_PN) (UseV walk_V))
LangEng: does John walk
LangSnd: و لی ٿا جان ھ ڇ
Lang: QuestVP whoSg_IP (UseV walk_V)
LangEng: who walks
LangSnd: و لی ٿير ھ ڪ
43
Lang: QuestIAdv why_IAdv (PredVP (UsePN john_PN) (UseV walk_V))
LangEng: why does John walk
LangSnd: و لی ٿو جان ھ ڇ
Lang: QuestIComp (CompIAdv where_IAdv) (UsePN john_PN)
LangEng: where is John
LangSnd: ی ی ا ھجان ڪٿ
Lang: IdetCN (IdetQuant which_IQuant (NumCard (NumNumeral (num (pot2as3 (pot1as2
(pot0as1 (pot0 n5)))))))) (UseN song_N)
LangEng: which five songs
LangSnd: ا پنج گانا ڪجي
Lang: IdetIP (IdetQuant which_IQuant (NumCard (NumNumeral (num (pot2as3 (pot1as2
(pot0as1 (pot0 n5))))))))
LangEng: which five
LangSnd: ا پنج ڪجي
Lang: IdetIP (IdetQuant which_IQuant NumSg)
LangEng: which
LangSnd: و ڪجي
Lang: PrepIP with_Prep whoSg_IP
LangEng: with whom
LangSnd: ڪھنجo سان
Lang: QuestIComp (CompIAdv where_IAdv) (UsePron it_Pron)
LangEng: where is it
LangSnd: ی ی ا ا ھا ڪٿ ھ
Lang: QuestIComp (CompIP whoSg_IP) (UsePron it_Pron)
LangEng: who is it
LangSnd: ی ير ا ا ھا ڪ ھ
44
Lang: ExistNP (DetCN (DetQuant IndefArt NumSg) (RelCN (UseN woman_N) (UseRCl
(TTAnt TPres ASimul) PPos (RelCl (PredVP (UsePN john_PN) (ComplSlash (SlashV2a
love_V2) (UsePron she_Pron)))))))
LangEng: there is a woman such that John loves her
LangSnd: ی و ا ری و سان عشق ا جان ھتی استری جي ٿ ڪ ھ ھڙ ھ
Lang: ExistNP (DetCN (DetQuant IndefArt NumSg) (RelCN (UseN woman_N) (UseRCl
(TTAnt TPres ASimul) PPos (RelVP IdRP (ComplSlash (SlashV2a love_V2) (UsePN
john_PN))))))
LangEng: there is a woman who loves John
LangSnd: ی ی ا ری ی جان سان عشق ھتی استری ج ٿ ڪ ھڙ ھ
Lang: ExistNP (DetCN (DetQuant IndefArt NumSg) (RelCN (UseN woman_N) (UseRCl
(TTAnt TPres ASimul) PPos (RelSlash IdRP (SlashVP (UsePN john_PN) (SlashV2a
love_V2))))))
LangEng: there is a woman whom John loves
LangSnd: ی و ا ری ن سان جان عشق ھتی استری جن ٿ ڪ ھ ھ
Lang: PredVP (UsePN john_PN) (UseV walk_V)
LangEng: John walks
LangSnd: و لی ٿجان ھ
Lang: RelCN (UseN girl_N) (UseRCl (TTAnt TPres ASimul) PPos (RelSlash IdRP
(SlashPrep (PredVP (UsePron he_Pron) (UseV walk_V)) with_Prep)))
LangEng: girl with whom he walks
LangSnd: و لی و ن سان ری جن ٿو ھ ھ ھ ڪ ڇ
Lang: ImpVP (ReflVP (SlashV2a love_V2))
LangEng: love yourself
LangSnd: ر پا سان ڻعشق ڪ
Lang: UseCl (TTAnt TCond AAnter) PNeg (PredVP (UsePN john_PN) (UseV walk_V))
45
LangEng: John wouldn't have walked
LangSnd: لندو ھجان شاید ن
Lang: UseQCl (TTAnt TCond AAnter) PNeg (QuestCl (PredVP (UsePN john_PN) (UseV
walk_V)))
LangEng: wouldn't John have walked
LangSnd: لندو ھا جان شاید ن ڇ
Lang: RelCN (UseN girl_N) (UseRCl (TTAnt TCond AAnter) PNeg (RelVP IdRP (UseV
walk_V)))
LangEng: girl who wouldn't have walked
LangSnd: لی ی شاید ن ری ج ھو ھڙ ڪ ڇ
Lang: RelCN (UseN girl_N) (UseRCl (TTAnt TCond AAnter) PNeg (RelSlash IdRP
(SlashPrep (PredVP (UsePron i_Pron) (UseV walk_V)) with_Prep)))
LangEng: girl with whom I wouldn't have walked
LangSnd: لون ن سان مان شاید ن ری جن ھو ھ ڪ ڇ
Lang: RelS (UseCl (TTAnt TPres ASimul) PPos (PredVP (UsePron she_Pron) (UseV
sleep_V))) (UseRCl (TTAnt TPres ASimul) PPos (RelVP IdRP (UseComp (CompAP (PositA
good_A)))))
LangEng: she sleeps , which is good
LangSnd: ی۶ھو و ا ا س ی ج ی ھ سم ٺ ھڙ ٿ ھ
Lang: TFullStop (PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PPos (PredVP
(UsePN john_PN) (UseV walk_V)))) NoVoc) TEmpty
LangEng: John walks .
LangSnd: و لی ٿجان ھ .
Lang: TQuestMark (PhrUtt NoPConj (UttQS (UseQCl (TTAnt TPres ASimul) PPos (QuestCl
(PredVP (UsePron they_Pron) (UseComp (CompAdv here_Adv)))))) NoVoc) TEmpty
LangEng: are they here ?
LangSnd: ن تی ا ی ھا ا ھ ھ ? ڇ
46
Lang: PredVP (UsePron i_Pron) (UseV sleep_V)
LangEng: I sleep
LangSnd: و ان ٿمان سم ھ
Lang: PredVP (UsePron i_Pron) (ComplVV want_VV (UseV run_V))
LangEng: I want to run
LangSnd: و ان و چا ٿمان ھ ڙڻ ڊ
Lang: PredVP (UsePron i_Pron) (ComplVS say_VS (UseCl (TTAnt TPres ASimul) PPos
(PredVP (UsePron she_Pron) (UseV run_V))))
LangEng: I say that she runs
LangSnd: و و ت ھمان چوان ی۶ٿ ی و ٿ ڙ ڊ
Lang: PredVP (UsePron i_Pron) (ComplVQ wonder_VQ (UseQCl (TTAnt TPres ASimul)
PPos (QuestVP whoSg_IP (UseV run_V))))
LangEng: I wonder who runs
LangSnd: و ی و ير و ت يان يران ٿمان ڙ ڊ ڪ ٿ ٿ ھ
Lang: PredVP (UsePron they_Pron) (ComplVA become_VA (PositA red_A))
LangEng: they become red
LangSnd: ا ين ا ا ی ٿا ٿ ڙھ ڳ ھ
Lang: PredVP (UsePron i_Pron) (ComplSlash (SlashV2S answer_V2S (UseCl (TTAnt TPres
ASimul) PPos (PredVP (UsePron it_Pron) (UseComp (CompAP (PositA good_A))))))
(UsePron he_Pron))
LangEng: I answer to him that it is good
LangSnd: ی و ا ا س و ت ا یان و کی جواب ھمان ٺ ھ ٿ ڏ ھ
Lang: PredVP (UsePron i_Pron) (ComplSlash (SlashV2A paint_V2A (PositA red_A))
(UsePron it_Pron))
LangEng: I paint it red
LangSnd: و ران و رنگ ا ن کی ٿمان ڪ ڙھ ڳ ھ
47
Lang: RelCN (UseN car_N) (UseRCl (TTAnt TPres ASimul) PPos (RelSlash IdRP (SlashVP
(UsePron i_Pron) (SlashVV want_VV (SlashV2a buy_V2)))))
LangEng: car which I want to buy
LangSnd: و ان ر چا ن مان ئرید ی جن ٿگا ھ ڻ ڪ ھ ڏ
Lang: PredVP (UsePron he_Pron) (ReflVP (SlashV2a love_V2))
LangEng: he loves himself
LangSnd: و ری ٿو پا سان عشق ڪ ڻ ھ
Lang: PredVP (DetNP (DetQuant this_Quant NumSg)) (UseComp (CompAP (PositA
warm_A)))
LangEng: this is warm
LangSnd: ی ھی گرم ا ھ
Lang: PredVP (UsePron we_Pron) (PassV2 love_V2)
LangEng: we are loved
LangSnd: ا رون ٿاسان عشق ڪ
Lang: PredVP (UsePron we_Pron) (AdvVP (UseV sleep_V) here_Adv)
LangEng: we sleep here
LangSnd: ا ون تی سم ٿاسان ھ ھ
Lang: PredVP (UsePron we_Pron) (AdVVP always_AdV (UseV sleep_V))
LangEng: we always sleep
LangSnd: ا ون ميش سم ٿاسان ھ ھ ھ
Lang: PredVP (UsePron we_Pron) (UseComp (CompAP (PositA small_A)))
LangEng: we are small
LangSnd: يون ا ا ھاسان نن ڍ
Lang: PredVP (UsePron i_Pron) (UseComp (CompNP (DetCN (DetQuant IndefArt NumSg)
(UseN man_N))))
LangEng: I am a man
LangSnd: يان و ا ھمان ما ڻھ
48
Lang: PredVP (UsePron i_Pron) (UseComp (CompAdv here_Adv))
LangEng: I am here
LangSnd: يان تی ا ھمان ھ
Lang: PhrUtt NoPConj (UttImpSg PPos (ImpVP (ComplSlash (SlashV2A paint_V2A
(ConjAP both7and_DConj (BaseAP (ComparA small_A (DetCN (DetQuant DefArt NumSg)
(UseN sun_N))) (ComparA big_A (DetCN (DetQuant DefArt NumSg) (UseN moon_N))))))
(DetCN (DetQuant DefArt NumSg) (UseN earth_N))))) NoVoc
LangEng: paint the earth both smaller than the sun and bigger than the moon
LangSnd: ر ٻرنگ و۶ڪ و ہ چن کان و ڏی سج کان نن ڊ ڍ
Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPres ASimul) PPos (PredVP (DetCN
every_Det (UseN baby_N)) (UseComp (CompNP (ConjNP either7or_DConj (BaseNP
(DetCN (DetQuant IndefArt NumSg) (UseN boy_N)) (DetCN (DetQuant IndefArt NumSg)
(UseN girl_N))))))))) NoVoc
LangEng: every baby is either a boy or a girl
LangSnd: و ار ڪر ٻ ھڪ ی۶ھ ری ا و رو یا يو چو ھی ڪ ڇ ڪ ٻ
Lang: PhrUtt NoPConj (UttAdv (ConjAdv either7or_DConj (ConsAdv here7from_Adv
(BaseAdv there_Adv everywhere_Adv)))) NoVoc
LangEng: either from here , there or everywhere
LangSnd: ن۶ڪو ر ن , اتی یا ان ي يو ڌی ھ ھ ھ ڏ ھ ٻ
Lang: RelCN (UseN bird_N) (UseRCl (TTAnt TPres ASimul) PPos (RelSlash IdRP (SlashVP
(UsePron i_Pron) (SlashVV want_VV (SlashV2A paint_V2A (PositA red_A))))))
LangEng: bird which I want to paint red
LangSnd: و ان و چا ا ر ن کی مان رنگ ٿپکی جن ھ ڙھ ڳ ڻ ڪ ھ
Lang: UttImpSg PPos (ImpVP (ComplVV want_VV (ComplSlash (SlashV2a buy_V2)
(UsePron it_Pron))))
LangEng: want to buy it
LangSnd: ر ڻچا ئرید ڪ ھ
49
Lang: UttImpSg PPos (ImpVP (ComplVV want_VV (ComplSlash (SlashV2A paint_V2A
(PositA red_A)) (UsePron it_Pron))))
LangEng: want to paint it red
LangSnd: ر و رنگ ا ڻچا ڪ ڙھ ڳ ھ
Lang: PhrUtt NoPConj (UttImpSg PPos (ImpVP (ReflVP (SlashV2Q ask_V2Q (UseQCl
(TTAnt TPast ASimul) PPos (QuestVP whoSg_IP (UseV come_V))))))) NoVoc
LangEng: ask yourself who came
LangSnd: ڻپ پا کان ڇ
Lang: PhrUtt NoPConj (UttS (UseCl (TTAnt TPast ASimul) PPos (PredVP (UsePron i_Pron)
(ReflVP (SlashV2A paint_V2A (ComparA beautiful_A (UsePN john_PN))))))) NoVoc
LangEng: I painted myself more beautiful than John
LangSnd: يم رون ھمان پا کی جان کان پيارو رنگ ڪ ڻ
Lang: UseCl (TTAnt TPast AAnter) PPos (PredVP (UsePron she_Pron) (ComplSlash
(SlashV2a buy_V2) (DetCN (DetQuant IndefArt NumSg) (AdjCN (PositA red_A) (UseN
house_N)))))
LangEng: she had bought a red house
LangSnd: ریل ۶ھو ر ئرید و گ ا ھ ڪ ھ ڙھ ی۶ڳ
In the Sindhi Resource Grammar test some sentences which do not give fully correct
translation.
Lang: UttImpSg PNeg (ImpVP (ReflVP (SlashV2a love_V2)))
LangEng: don't love yourself
LangSnd: ر پا سان ڻن عشق ڪ
Lang: UttImpPl PNeg (ImpVP (ReflVP (SlashV2a love_V2)))
LangEng: don't love yourselves
LangSnd: ر پا سان ڻن عشق ڪ
Lang: QuestSlash whoSg_IP (SlashVP (UsePN john_PN) (SlashV2a love_V2))
LangEng: whom does John love
50
LangSnd: و ری ٿير جان عشق ڪ ڪ
Lang: PredVP (UsePron i_Pron) (ComplSlash (Slash3V3 give_V3 (UsePron he_Pron))
(UsePron it_Pron))
LangEng: I give it to him
LangSnd: و یان و ن ٿمان ڏ ھ ھ
Lang: RelCN (UseN girl_N) (UseRCl (TTAnt TPres ASimul) PPos (RelSlash IdRP (SlashVP
(UsePron he_Pron) (SlashV2a see_V2))))
LangEng: girl whom he sees
LangSnd: و سی و ن ری جن ٿو ڏ ھ ھ ڪ ڇ
Lang: RelCN (UseN girl_N) (UseRCl (TTAnt TPres ASimul) PPos (RelSlash IdRP
(AdvSlash (SlashVP (UsePron he_Pron) (SlashV2a see_V2)) today_Adv)))
LangEng: girl whom he sees today
LangSnd: و سی و ن ا ری جن ٿو ڏ ھ ڄ ھ ڪ ڇ
Lang: RelCN (UseN girl_N) (UseRCl (TTAnt TPres ASimul) PPos (RelSlash IdRP (SlashVS
(UsePron she_Pron) say_VS (UseSlash (TTAnt TPres ASimul) PPos (SlashVP (UsePron
he_Pron) (SlashV2a love_V2))))))
LangEng: girl whom she says that he loves
LangSnd: و ن سان ری جن ھو ھ ڪ و۶ڇ ری و عشق ی ت ٿ چوی ڪ ھ ٿ
51